Acta Neuropsychiatrica

BackgroundMajor depressive disorder (MDD) and its most severe phenotype, major dysmood disorder (MDMD), are distinguished by the activation of the immune-inflammatory response system, T cell activation, and a relative T regulatory cell suppression. Nevertheless, these immune data were not used to characterize the features of the immune protein-protein interaction (PPI) network of MDMD. ObjectivesTo identify the networks nodes and bottlenecks as well as the biological processes that are overrepresented in the PPI network, we conducted PPI network, annotation, and enrichment analyses. ResultsThe PPI network analysis has identified the following backbone genes: tumor necrosis factor- (TNF), interleukin (IL)6, CXCL12, CXCL10, CCL5, cluster of differentiation (CD)4, CD8A, human leukocyte antigen (HLA)-DR, and FOXP3. A "cellular and defense response", an "immune response system response", and "a viral process that involves viral protein interaction with cytokines and cytokine receptors" were all highly associated with the network. The chemokine network and TNF and nuclear factor-{kappa}B (NFKB) pathways are additional biological pathways that are enriched in the PPI network. Molecular complex detection extracted one component from the data, including viral protein interaction with cytokine and cytokine receptors and "regulated by RELA" (an NFKB subunit). ConclusionsViral processes may underlie the activation of T cells and the cytokine and chemokine networks that are associated with MDMD. Future research on the pathogenesis of MDMD and MDD should examine whether and which viral infections are associated with the onset of these conditions, or whether viral reactivation is associated with the recurrence of illness.

Understanding the alterations in brain function across different episodes of bipolar disorder (BD), including manic (BipM), depressive (BipD), and remission states (rBD), poses a significant challenge. In our cross-sectional study, we collected resting-state functional magnetic resonance imaging data from 117 BD patients (BipM: 38, BipD: 42, rBD: 37) and 35 healthy controls. Our aim was to delineate functional connections associated with episode dynamics, delineate common and specific patterns, validate them as biomarkers, and elucidate their biological underpinnings. Initially, we identified a common altered pattern within the subregions of the ventral-attention network, alongside specific patterns observed in the default mode network for BipM, the prefrontal network for BipD, and the limbic network for rBD. Using large-sample data from the Human Connectome Project, we further identified that these connectivity patterns exhibit relatively high reliability and heritability. Also, these distinct patterns accurately characterized the diverse episodes of BD and effectively predicted the corresponding clinical symptoms linked with each episode type. Importantly, using out of sample data to decode possible neurobiological mechanisms underlying these patterns, we found that regions of particular interest were enriched in multiple receptors, including MOR, NMDA, and H3 for specific alterations, and A4B2, 5HTT, and 5HT1a for common alterations. Moreover, both episode-specific and common patterns demonstrated a high enrichment for cell types such as L5ET, Micro/PVM,oligodendrites and Chandelier. Our study offers novel insights concerning episode dynamics in BD, paving the way for personalized medicine approaches tailored to address the various episodes.

BackgroundCritical COVID-19 disease is accompanied by depletion of plasma tryptophan (TRY) and increases in indoleamine-dioxygenase (IDO)-stimulated production of neuroactive tryptophan catabolites (TRYCATs), including kynurenine (KYN) and quinolinic acid. The TRYCAT pathway has not been studied extensively in association with the physiosomatic and affective symptoms of Long COVID. MethodsIn the present study, we measured serum tryptophan (TRY), TRYCATs, insulin resistance (using the HOMA2-IR index), C-reactive protein (CRP), physiosomatic, depression and anxiety symptoms in 90 Long COVID patients, 3-10 months after remission of acute infection. ResultsWe were able to construct an endophenotypic class of severe Long COVID (22% of the patients) with very low TRY and oxygen saturation (SpO2, during acute infection), increased kynurenine, KYN/TRY ratio, CRP, and very high ratings on all symptom domains. One factor could be extracted from physiosomatic symptoms (including chronic fatigue-fibromyalgia), depression, and anxiety symptoms, indicating that all domains are manifestations of the common physio-affective phenome. Three Long COVID biomarkers (CRP, KYN/TRY, IR) explained around 40% of the variance in the physio-affective phenome. The latter and the KYN/TRY ratio were significantly predicted by peak body temperature (PBT) and lowered SpO2 during acute infection. One validated latent vector could be extracted from the three symptom domains and a composite based on CRP, KYN/TRY, IR (Long COVID), and PBT and SpO2 (acute COVID-19). ConclusionThe physio-affective phenome of Long COVID is a manifestation of inflammatory responses during acute and Long COVID and lowered plasma tryptophan and increased kynurenine may contribute to these effects.

ObjectivePediatric acute-onset neuropsychiatric syndrome (PANS) is a complex neuropsychiatric syndrome characterized by an abrupt onset of obsessive-compulsive symptoms and/or severe eating restrictions, along with at least two concomitant debilitating cognitive, behavioral, or neurological symptoms. A wide range of pharmacological interventions along with behavioral and environmental modifications, and psychotherapies have been adopted to treat symptoms and underlying etiologies. Our goal was to develop a data-driven approach to identify treatment patterns in this cohort. Materials and MethodsIn this cohort study, we extracted medical prescription histories from electronic health records. We developed a modified dynamic programming approach to perform global alignment of those medication histories. Our approach is unique since it considers time gaps in prescription patterns as part of the similarity strategy. ResultsThis study included 43 consecutive new-onset pre-pubertal patients who had at least 3 clinic visits. Our algorithm identified six clusters with distinct medication usage history which may represent clinicians practice of treating PANS of different severities and etiologies i.e., two most severe groups requiring high dose intravenous steroids; two arthritic or inflammatory groups requiring prolonged nonsteroidal anti-inflammatory drug (NSAID); and two mild relapsing/remitting group treated with a short course of NSAID. The psychometric scores as outcomes in each cluster generally improved within the first two years. Discussion and conclusionOur algorithm shows potential to improve our knowledge of treatment patterns in the PANS cohort, while helping clinicians understand how patients respond to a combination of drugs.

BackgroundPatients with transfusion-dependent thalassemia (TDT) are vulnerable to neurotoxicity due to frequent blood transfusions and the subsequent iron overload (IO) and inflammation. As a result, affective (depression and anxiety) and chronic fatigue syndrome (CFS) symptoms may develop. AimsTo investigate the potential association between TDT and neuronal injury, as assessed with serum concentrations of neuronal damage biomarkers, including neurofilament light (NFL), glial fibrillary acidic protein (GFAP), neuron-specific enolase (NSE), and nestin. MethodsWe investigated the associations between those CNS injury biomarkers, neuro-immune markers (C-reactive protein (CRP), interleukin (IL)-6, and IL-10), calcium, magnesium, copper and zinc, and the Fibro-Fatigue (FF), the Childrens Depression Inventory (CDI), and the Spence Childrens Anxiety Scale (SCAS) scores in 126 children with TDT and 41 healthy children. ResultsTDT children show significant increases in IO, FF, CDI, and SCAS scores, serum NSE, GFAP, NF-L, CRP, copper, IL-6, and IL-10, and lowered magnesium, zinc, and calcium as compared with healthy children. There were significant correlations between the CDI score and NFL, NSE and GFAP; SCAS score and NFL, and FF score and NFL and GFAP. The neuronal damage biomarkers (except nestin) were significantly associated with inflammatory, erythron (hematocrit and hemoglobin) and IO (iron and ferritin) biomarkers. ConclusionsTDT is characterized by intertwined increases in neuronal injury biomarkers and neuropsychiatric symptoms suggesting that TDT-associated neurotoxicity plays a role in affective symptoms and CFS due to TDT. Inflammation and neurotoxicity are novel drug targets for the prevention of affective symptoms and CFS due to TDT.

Psychiatric disorders, including eating disorders (EDs), are characterized by heterogeneity that limits diagnostic accuracy and treatment personalization. Precision psychiatry calls for tools to identify data-driven phenotypes beyond traditional categories. In a large cohort (N=809), we applied an unsupervised clustering pipeline to self-report and clinical variables to uncover ED subgroups. Repeated Spectral Clustering revealed four phenotypes: two aligned with DSM-5 diagnoses (anorexia nervosa restricting-type; bulimia nervosa), and two diagnostically mixed clusters, one characterized by higher psychopathology and trauma exposure, the other by prolonged illness duration. These clusters showed distinct profiles and outcomes. The higher predictability of data-driven clusters compared to DSM-5 categories suggests that unsupervised stratification may offer a complementary perspective on relevant heterogeneity. Our findings highlight how computational phenotyping can advance precision psychiatry by revealing meaningful patterns, informing prognosis, and guiding personalized interventions. This approach may help bridge the gap between clinical presentations and the need for stratified, patient-centered care.

BackgroundThe term "mood stabilizer" is controversial in the literature. As there is no consensual meaning, its retirement has been suggested to avoid misuse. Nevertheless, it remains largely employed, and may carry an important meaning. This issue has not been approached using a validated qualitative inquiry. MethodsWe employed document analysis for reviewing definitions for mood stabilizer. Then, we used concept analysis as a qualitative methodology to clarify the meanings associated with the term. Based on its results, we built a theoretical model for a mood stabilizer, matching it with evidence for drugs used in the treatment of bipolar disorder. ResultsConcept analysis of documents defining the term unearthed four attributes of a mood stabilizer that were nested into the following ascending hierarchy: "not worsening", "acute effects", "prophylactic effects", and "advanced effects". To be considered a mood stabilizer, a drug had to reach the "prophylactic effects" tier, as this was discussed by authors as the core aspect of the class. After arranging drugs according to this scheme, "lithium" and "quetiapine" received the label, but only the former fulfilled all four attributes, as evidence indicates it has neuroprotective action. ConclusionThe proposed model uses a hierarchy of attributes that take into account the complexity of the term and help to determine whether a drug is a mood stabilizer. Prophylaxis is pivotal to the concept, whose utility lies in implying a drug able to truly treat bipolar disorder, as opposed to merely targeting symptoms. This could modify long-term outcomes and illness trajectory.

The first publication demonstrating that major depressive disorder (MDD) is associated with alterations in the gut microbiota appeared in 2008 (Maes et al., 2008). The purpose of the present study is to delineate a) the microbiome signature of the phenome of depression, including suicidal behaviours and cognitive deficits; the effects of adverse childhood experiences (ACE) and recurrence of illness index (ROI) on the microbiome; and the microbiome signature of lowered high-density lipoprotein cholesterol (HDLc). We determined isometric log-ratio abundances or prevalence of gut microbiome phyla, genera, and species by analyzing stool samples from 37 healthy Thai controls and 32 MDD patients using 16S rDNA sequencing. Six microbiome taxa accounted for 36% of the variance in the depression phenome, namely Hungatella and Fusicatenibacter (positive associations) and Butyricicoccus, Clostridium, Parabacteroides merdae, and Desulfovibrio piger (inverse association). This profile (labeled enterotype 1) indicates compositional dysbiosis, is strongly predicted by ACE and ROI, and is linked to suicidal behaviours. A second enterotype was developed that predicted a decrease in HDLc and an increase in the atherogenic index of plasma (Bifidobacterium, P. merdae, and Romboutsia were positively associated, while Proteobacteria and Clostridium sensu stricto were negatively associated). Together, enterotypes 1 and 2 explained 40.4% of the variance in the depression phenome, and enterotype 1 in conjunction with HDLc explained 39.9% of the variance in current suicidal behaviours. In conclusion, the microimmuneoxysome is a potential new drug target for the treatment of severe depression and suicidal behaviours, and possibly for the prevention of future episodes.

Bipolar disorder (BD) is a chronic psychiatric condition affecting approximately 1-2% of the population, characterized by depressive and manic episodes. BD comprises two main subtypes, defined by the presence of mania (BD-I) or hypomania (BD-II). Commonly used clinical scales, including the Global Assessment of Functioning (GAF), Clinical Global Impressions (CGI), and World Health Organization Disability Assessment Schedule (WHODAS), assess functional impairment at the time of evaluation. However, they may not adequately capture cumulative lifetime illness burden or provide a retrospective measure of clinical severity. Here, we introduce the Index of Number of Events and Severity (INES), a novel instrument designed to quantify longitudinal illness-course severity in BD by integrating cumulative clinical events with illness duration. INES incorporates psychosis and rapid cycling as dichotomous variables and quantifies hospitalizations, suicide attempts, and affective episodes as discrete categories. INES was evaluated in 307 individuals from the MadManic cohort. It correlated moderately with GAF and CGI, while its strongest association was observed with WHODAS (r=0.347). Factor analysis over the four scales supported a two-factor structure, where INES loaded alongside WHODAS, capturing the variability of structured instruments. Linear modelling indicated that traditional scales explained only 14.4% of the variance of INES, suggesting that this scale captures clinical information largely unaccounted by the other instruments. INES was the only to differentiate between BD subtypes, with higher severity observed in individuals with BD-I. These findings support INES as a reproducible tool for capturing cumulative lifetime severity in BD, with potential utility in clinical and genetic studies.

BackgroundMajor depressive disorder (MDD) is a neuro-immune, oxidative, and nitrosative stress (NIMETOX) disorder, in which peripheral immune-redox pathways intersect with metabolic networks leading to neurotoxicity within the limbic-prefrontal affective circuits. Comprehensive metabolomics analysis in well-phenotyped patients is vital to elucidate their metabolic profile. ObjectivesTo identify metabolic abnormalities that differentiate inpatients with severe MDD from healthy controls through high-resolution, untargeted metabolomics. MethodsSerum samples from 125 MDD inpatients and 40 healthy controls were analyzed utilizing liquid chromatography and mass spectrometry. A meticulously regulated multistage machine learning pipeline with leakage-prevention protocols was employed to analyze differences between MDD and controls and to predict phenome scores. ResultsFeature selection showed that 16 metabolites and 6 functional modules reliably distinguished MDD. The functional profile of the metabolites indicates a convergence of lipotoxicity, phospholipid remodeling, disruptions in fatty acid metabolism, mitochondrial redox imbalance, ether-lipid metabolism, and antioxidant depletion. This MDD metabotype was not affected by metabolic syndrome. A substantial portion of the variance in overall depression severity (72.5%), physiosomatic symptoms (55.8%) and suicidal ideation (23.6%) was accounted for by increased lipitoxicity, phospholipid remodeling, and fatty acid storage/signaling. The recurrence of illness (27.7%) was associated with a self-reinforcing-lipid-redox-inflammatory module that maintains cellular stress. DiscussionThe MDD metabotype represents a cohesive metabolic network that is associated with the NIMETOX pathogenesis of MDD. Metabolomics provides a comprehensive foundation for subtyping and precision psychiatry. Lipoxygenase-15, lipotoxicity, phospholipase A2, and lipid-redox intersections are important drug targets to treat MDD.

IntroductionCommon diseases are influenced by a variety of factors that can enhance one persons susceptibility to developing a specific condition. Complex traits have been investigated in several biological levels. One that reflects the high interconnectivity and interaction of genes, proteins and transcription factors is the transcriptome. In this study, we disclose the protocol for a systematic review and meta-analysis aiming at summarizing the available evidence regarding transcriptomic gene expression levels of peripheral blood samples comparing subjects with psychiatric, neurological and other common disorders to healthy controls. Methods and analysisThe investigation of the transcriptomic levels in the peripheral blood enables the unique opportunity to unravel the etiology of common diseases in patients ex-vivo. However, the experimental results should be minimally consistent across studies for them to be considered as the best approximation of the true effect. In order to test this, we will systematically identify all transcriptome studies that compared subjects with common disorders to their respective control samples. We will apply meta-analyses to assess the overall differentially expressed genes throughout the studies of each condition. Ethics and disseminationThe data that will be used to conduct this study are available online and have already been published following their own ethical laws. Therefore this study requires no further ethical approval. The results of this study will be published in leading peer-reviewed journals of the area and also presented at relevant national and international conferences. Strengths and limitations of this study We present a new and systematically centered method to assess the overall effect of transcriptomic levels in the blood of subjects with common conditions. Meta-analyses are a robust statistical method to assess effect sizes across studies. The analysis is limited by the availability of studies, as well as their quality and comprehensiveness. Subgroup and meta-regression analyses will be also limited by the amount and quality of sample characterization variables made available by original studies.

BackgroundMajor depression comprises two discrete subtypes, major (MDMD) and simple (SDMD) dysmood disorder. MDMD, but not SDMD, patients were identified to have highly sensitized cytokine/growth factor networks using stimulated whole blood cultures. However, no information regarding serum cytokines/chemokines/growth factors in SDMD is available. ObjectivesThis case-control study compares 48 serum cytokines/chemokines/growth factors in academic students with SDMD (n=64) and first episode (FE)-SDMD (n=47) to those of control students (n=44) using a multiplex assay. FindingsBoth FE-SDMD and SDMD exhibit a notable inhibition of immune profiles, such as the compensatory immunoregulatory response system (CIRS) and alternative M2 macrophage and T helper-2 (Th-2) profiles. We observed a substantial reduction in the serum concentrations of five proteins: interleukin (IL)-4, IL-10, soluble IL-2 receptor (sIL-2R), IL-12p40, and macrophage colony-stimulating factor. A significant proportion of the variability observed in suicidal behaviors (26.7%) can be accounted for by serum IL-4, IL-10, and sIL-2R (all decreased), and CCL11 (eotaxin) and granulocyte CSF (both increased). The same biomarkers (except for IL-10), accounted for 25.5% of the variance in SDMS severity. A significant correlation exists between decreased levels of IL-4 and elevated ratings of the brooding type of rumination. ConclusionsThe immune profile of SDMD and FE-SDMD exhibits a significant deviation from that observed in MDMD, providing additional evidence that SDMD and MDMD represent distinct phenotypes. SDMD is characterized by the suppression of the CIRS profile, which signifies a disruption of immune homeostasis and tolerance, rather than the presence of an inflammatory response.

ObjectiveMood disorder (including major depression and bipolar disorder) prevalence is over 10% and accounts for a significant share of global disease burden. Mental and physical illness are related, however, the association between mood disorders and acute/chronic disease subclasses remains poorly understood. MethodsThis observational cross-sectional study used administrative claims data from 6,709,258 adult enrollees with a full-year enrollment in the 2018 OPTUM Clinformatics(R) database. Data of enrollees with/without diagnoses of a mood disorder co-occurring with chronic comorbid conditions (defined by the Elixhauser Comorbidity Index) using the International Classification of Diseases (ICD-10) were analyzed by accounting for age, race, and ethnicity. ResultsOverall, the sample was predominantly non-Hispanic Caucasians (64.56%), with 48.59% females and a mean age of 43.54 years{+/-}12.46 years. The prevalence of mood disorders was 3.71% (248,890), of which 0.61% (n=40,616) had bipolar disorders and 3.10% (n=208,274) had Major Depressive Disorder (MDD). Logistic regression odds ratios revealed a strong association between mood disorder diagnoses and peptic ulcers (2.11; CI=2.01-2.21), weight loss (2.53; CI, 2.46-2.61), renal failure (2.37, CI = 2.31-2.42), peripheral vascular disease (2.24; CI=2.19-2.30), and pulmonary circulation disorder (1.77; CI=1.70-1.84). ConclusionsOverall, mood disorders were associated with vascular and cardiac chronic medical conditions, suggesting a possible pathophysiological link between these conditions. The results highlight the importance of understanding the prevalence of co-occurring mood and medical conditions and may inform novel biological diagnostics and future identification of mechanisms for multimorbidity.

BackgroundMajor depressive disorder (MDD) is accompanied by activated neuro-immune pathways, increased physiosomatic and chronic fatigue-fibromyalgia (FF) symptoms. The most severe MDD phenotype, namely major dysmood disorder (MDMD), is associated with adverse childhood experiences (ACEs) and negative life events (NLEs) which induce cytokines/chemokines/growth factors. AimsTo delineate the impact of ACE+NLEs on physiosomatic and FF symptoms in first episode (FE)-MDMD, and examine whether these effects are mediated by immune profiles. MethodsACEs, NLEs, physiosomatic and FF symptoms, and 48 cytokines/chemokines/growth factors were measured in 64 FE-MDMD patients and 32 normal controls. ResultsPhysiosomatic, FF and gastro-intestinal symptoms belong to the same factor as depression, anxiety, melancholia, and insomnia. The first factor extracted from these seven domains is labeled the physio-affective phenome of depression. A part (59.0%) of the variance in physiosomatic symptoms is explained by the independent effects of interleukin (IL)-16 and IL-8 (positively), CCL3 and IL-1 receptor antagonist (inversely correlated). A part (46.5%) of the variance in physiosomatic (59.0%) symptoms is explained by the independent effects of interleukin (IL)-16, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) (positively) and combined activities of negative immunoregulatory cytokines (inversely associated). Partial Least Squares analysis shows that ACE+NLEs exert a substantial influence on the physio-affective phenome which are partly mediated by an immune network composed of IL-16, CCL27, TRAIL, macrophage-colony stimulating factor, and stem cell growth factor. ConclusionsThe physiosomatic and FF symptoms of FE-MDMD are partly caused by immune-associated neurotoxicity due to T helper (Th)-1 polarization, Th-1, and M1 macrophage activation and relative lowered compensatory immunoregulatory protection.

COVID-19 infections in adults often result in medical, neuropsychiatric, and unspecific symptoms, called Long COVID, and the premorbid functional status cannot be achieved. Regarding the course in children and adolescents, however, reliable data are not yet available. Objective380 children and adolescents/young adults aged between 6 and 21 years, being treated for various psychiatric diseases in an outpatient clinical service, were examined for COVID-19 infections and Long COVID symptoms following a structured protocol. ResultsThree patients had COVID-19; one patient had symptoms of Long COVID in his medical history, but they could not be objectivized in an in-depth neuropsychiatric and neuropsychological assessment. ConclusionsLong COVID seems to occur rarely in children and adolescents. Objectivizing the symptoms is a difficult task that requires various diagnostic considerations.

Depression and anxiety are highly correlated, yet little is known about the course of each condition when presenting concurrently. This study aimed to identify longitudinal patterns and changes in depression and anxiety symptoms during antidepressant treatment, and evaluate clinical factors associated with each response pattern. Self-report Patient Health Questionnaire-9 (PHQ-9) and General Anxiety Disorder-7 (GAD-7) scores were used to track the courses of depression and anxiety respectively over a three-month window, and group-based trajectory modeling was used to derive subgroups of patients who have similar response patterns. Multinomial regression was used to associate various clinical variables with trajectory subgroup membership. Of the 577 included adults, 373 (64.6%) were women, and the mean age was 39.3 (SD: 12.9) years. Six depression and six anxiety trajectory subgroups were computationally derived; three depression subgroups demonstrated symptom improvement, and three exhibited nonresponse. Similar patterns were observed in the six anxiety subgroups. Factors associated with treatment nonresponse included higher pretreatment depression and anxiety severity and poorer sleep quality, while better overall health and younger age were associated with higher rates of remission. Synchronous and asynchronous paths to improvement were also observed between depression and anxiety. High baseline depression or anxiety severity alone may be an insufficient predictor of treatment nonresponse. These findings have the potential to motivate clinical strategies aimed at treating depression and anxiety simultaneously.

Background: Personality traits are consistently associated with bipolar disorder (BD). However, their features across BD diagnostic subtypes and their modulation by demographic and health-related factors remain largely unexplored. This study aimed to characterize Big Five personality domains in individuals with BD compared to controls, and to examine differences between BD type I (BD-I) and BD type II (BD-II). Methods: We analyzed 833 participants from the MadManic cohort (300 BD subjects and 533 controls) with available Big Five Inventory-2 (BFI-2) data. Linear regression models were used to assess associations between personality traits and BD diagnosis, adjusting for relevant covariates. Additional comparisons were conducted across sex, age, and Body Mass Index (BMI), and between BD-I and BD-II patients. Results: BD was associated with higher Negative Emotionality (NE) and lower Extraversion and Conscientiousness. Conscientiousness was also inversely associated with BMI. Within the BD group, individuals with BD-I exhibited lower NE compared to those with BD-II. Stratified analyses indicated that elevated NE in BD was the most consistent domain across sex, age, and BMI subgroups, whereas differences in Extraversion and Conscientiousness varied depending on subgroup features. Conclusions: BD is characterized by a distinct personality profile marked by elevated NE and reduced Extraversion and Conscientiousness. NE emerged as the most robust domain associated with BD, which may also differentiate between subtypes, with higher levels observed in BD-II than BD-I. These findings highlight the relevance for considering demographic and health-related factors, particularly BMI, when interpreting personality patterns in BD, supporting the role of personality dimensions to examine clinical heterogeneity.

The authors have withdrawn this manuscript because the results were posted in error. The authors do not wish this work to be cited as reference for the project. Please contact the corresponding author if you have any questions.

BackgroundBipolar disorder (BD) is frequently underdiagnosed, particularly in patients presenting with depressive disorders, leading to delays in appropriate treatment. Artificial intelligence (AI) applied to electronic health records (EHRs) may improve early detection by identifying clinically relevant symptom patterns. ObjectiveTo evaluate the diagnostic performance of a natural language processing (NLP)-based AI model for detecting BD-related features in EHRs of patients with affective diagnoses. MethodsA retrospective diagnostic accuracy study was conducted using 500 EHRs from a psychiatric referral hospital in Bogota, Colombia (2020-2024). The model extracted 18 predefined clinical domains from unstructured text and classified patients into four risk categories. Diagnostic performance was assessed in a validation subset of 100 records using independent psychiatric evaluation as the reference standard. Sensitivity, specificity, positive and negative predictive values, F1-score, and area under the receiver operating characteristic curve (AUC-ROC) were calculated. ResultsThe model achieved high agreement in symptom extraction (mean 91.1%). Sensitivity was 96.4% (95% CI: 87.7%-99.0%) and specificity was 84.4% (95% CI: 71.2%-92.3%), with an F1-score of 0.92 and an AUC-ROC of 0.932 (95% CI: 0.881-0.975). A substantial proportion of patients with depressive diagnoses were identified as having confirmed BD or clinically relevant risk. The model analyzed complete EHRs 120 times faster than human reviewers. ConclusionsNLP-based analysis of EHRs can achieve clinically meaningful performance in identifying BD-related patterns while substantially reducing review time. The model may be useful as a clinical decision support tool for earlier identification of bipolar disorder.

BackgroundCognitive deficits in bipolar affective disorder (BPAD), particularly during manic episodes, are well-documented. However, research on domain-specific memory impairments in bipolar mania is limited, especially in the Indian subcontinent. This study aimed to assess memory impairments in patients with bipolar mania using the Postgraduate Institute Memory (PGI-Memory) Scale and to highlight domain-specific deficits compared to healthy controls. MethodsThis cross-sectional study was conducted at Tertiary Care Centre in North India. Twenty patients diagnosed with bipolar mania and twenty age, sex and education-matched healthy controls between age 18 to 40 were recruited. Memory functions were assessed using the PGI-Memory scale, focusing on immediate, recent, remote, long-term memory, and associative memory. Mental control and working memory were also evaluated. ResultsBoth groups were matched in terms of age, sex, and education. the mean (sd) age for bipolar mania group was 27.2 (4.14) years. Patients with bipolar mania demonstrated significant deficits in various memory domains, including immediate, recent, remote, long-term, and associative memory, as well as in visual reproduction and recognition tasks. In contrast, their working memory performance was comparable to that of the control group. The largest deficits were observed in long-term memory (d =2.37) and visual reproduction (d=2.30). ConclusionsBipolar mania is associated with widespread memory impairments, particularly in long-term and associative memory, which may contribute to difficulties in emotional regulation and daily functioning. These findings emphasize the importance of considering memory impairments in the diagnosis and management of BPAD. Further studies are required to investigate the neurobiological foundations of these impairments and to develop specific interventions.